ArvicanthisDivision. Morphological traits place Rhabdomys in a group containing species of Arvicanthis, Lemniscomys, Mylomys, and Pelomys (Musser, 1987b), which is corroborated by analysis of mitochondrial gene sequences (cytochrome b, 12S and 16S rRNA gene fragments), and also includes Desmomys (Ducroz et al., 2001). The sequence data of Ducroz et al. also indicate that within this arvicanthine cluster, Rhabdomys and Desmomys are members of a lineage separate from that comprised of only Lemniscomys, and from another containing Arvicanthis, Mylomys, and Pelomys. Analysis of microcomplement fixation of albumin groups Rhabdomys with Lemniscomys, Pelomys, Grammomys, and Thallomys (Watts and Baverstock, 1995a).

When Wroughton (1905b) reviewed R. pumilio, he distinguished four morphological groups, each with different forms, and although unsure about the taxonomic status to give the forms, thought each group represented a separate species. Distribution of character variation, however, forced him to conclude that (p. 630) "in view of the absolute identity of pattern, the variability of coloration, and the difficulty of deciding the inter-relationship of the different forms, the simpler and safer way is to call them all subspecies of the original species pumilio." Wroughton's view prevails today. Checklists (G. M. Allen, 1939; Ellerman, 1941; Ellerman et al., 1953), faunal studies (e.g., Ansell, 1978; Ansell and Dowsett, 1988; Roberts, 1951; Skinner and Smithers, 1990; Smithers, 1971), a study of possible influences of climate on length of tail (Coetzee, 1970), and preliminary studies on geographic variation (see Meester et al., 1986:275) have not critically analyzed patterns of variation in pelage coloration and pattern along with morphology to assess whether only one or several species exist. Interpopulation breeding studies (Pillay, 2000) and allozymic analyses (Mahida et al., 1999) of South African populations have provided new comparative data, but these results are ambiguous in ascertaining whether more than one species is represented among the samples. Analysis of the allozymic data resulted in ranges of genetic similarities suggesting different subspecies were involved, if supported by other evidence, which Mahida et al. (1999) promised to provide derived from DNA sequencing and cytogenetic studies.

Analyses by Rambau et al. (2003;564) have provided that evidence in the form of mtDNA cytochrome b sequences and cytogenetic data. Two major and highly divergent lineages were identified, one inhabiting mesic regions of southern Africa, characterized by dark gray or dark brown fur and 2n = 46 or 48. The other occupies xeric habitats, and has much paler pelage and 2n = 48. High genetic distance, ecological difference, and differences in mating behavior between the two lineages indicate two species: R. pumilio in xeric grasslands, R. dilectus living in mesic environments. Origin of the two may have been facilitated by survival of an ancestral population in two refugia during climatic oscillations in the Pliocene, one in the mesic regions of NE Africa, the other in drier environments of southern Africa. Two species had already been recognized in Angola by Hill and Carter (1941:102); R. pumilio from the C and S highlands, and R. bechuanae form the arid southwestern desert just north of Namibia. They noted the lack of intergradation between the two kinds and we have not found any evidence in their material (in AMNH) that such intergradation exists. The samples from the highlands are likely P. dilecuts, the desert sample is P. pumilio (see accounts below).

Several studies are pertinent to systematics of Rhabdomys, but were published before the resolution of two species. Morphology of digestive system in relation to diet and evolution described by Perrin and Curtis (1980). In a study of size variation (using greatest length of skull) among samples from across southern Africa, Yom-Tov (1993) discovered a positive correlation between size and mean minimum temperature of the coldest month. He also determined that samples of Rhabdomys from the zone of sympatry with the Lemniscomys griselda complex, which are also diurnal herbivores but much larger in body size, are significantly smaller than samples from regions in which Lemniscomys does not occur. Yom-Tov suggested character release to be a primary factor in determining body size of Rhabdomys in southern Africa. Spermatozoal morphology documented by Breed (1995a). Populations in Southern African Subregion reviewed by de Graaff (1981, 1997e) and Skinner and Smithers (1990).

Rhabdomys is represented by isolated molars from the Pleistocene of Namibia (Senut et al., 1992); early Pleistocene Swartkrans and Sterkfontein cave sediments in South Africa (Avery, 1998, 2000). The earliest record is from Langebaanweg in South Africa and may be five million years old (early Pliocene; Denys, 1999).